Temperature Effects in Anaerobic Digestion Modeling
Wenche Hennie Bergland Carlos Dinamarca
Department of Process, Energy and Environmental Technology, Telemark University College, Norway,
Wenche.Bergland@hit.no, Carlos.Dinamarca@hit.no, Rune.Bakke@hit.no
Abstract
Temperature effects on kinetic coefficients for the
biochemical processes particle disintegration,
hydrolysis and substrate uptake reactions were included
in the anaerobic digestion model 1 (ADM1). It was
evaluated on data from a pilot experiments in a 220 liter
AD sludge bed reactor treating diary manure for 4
months of varying loads and temperatures; 25, 30 and
35 °C. Implementing individual temperature effects for
each biochemical reaction gave the best fit for both
biogas production and intermediate products. Simulated
overall soluble and particulate organic carbon removal,
pH and acetate are close to measured values while
propionate is underestimated. Temperature has a
moderate influence on steady state biogas production in
sludge bed AD (1.6 % per degree at 30 – 35 °C and 3.4
% per degree at 25 – 30 °C), implying the net energy
gain can peak at T < 35 °C in some cases.
Keywords: Anaerobic digestion, sludge bed,
temperature dependence, ADM1.
1 Introduction
Anaerobic digestion (AD) can be used to recover energy
as methane from organic wastes and thereby reduce
greenhouse gas emissions by contributing to more
sustainable waste handling.
Anaerobic degradation of organic matter into biogas
is carried out by a consortium of microorganisms, which
degrade complex organic macro-molecules by
extracellular (disintegration, hydrolysis) and
intracellular (acidogenesis, acetogenesis,
methanogenesis) enzyme mediated biochemical
reactions (Fig. 1). The Anaerobic Digestion Model No.1
(ADM1) (Batstone et al., 2002) is a common platform
of modelling, simulations and understanding AD. It was
developed by the International Water Association
(IWA).
ADM1 was developed primarily to model digestion
of sludge from wastewater treatment plants where the
standard process temperatures are 35 °C, or 55 °C,
presumed optimal for respectively meso- and
thermophilic digestion. Energy optimized AD may,
however, not be such, e.g. 35 °C for mesophilic. Higher
temperatures (35 - 42 °C) are sometimes preferred while
lower digestion temperature than microbial optima for
which standard model parameters are available may also
DOI Proceedings of the 56th SIMS
10.3384/ecp15119261 October 07-09, 2015, Linköping, Sweden
Rune Bakke
be favorable in some cases. Anaerobic reactors fed with
low energy substrates, such as cow manure slurry, can
possibly be an example of the latter since a large portion
of the potential energy might be required to heat the
reactor to 35 °C in cold climates. The potential for
reducing heating requirements is therefore examined by
testing and modeling AD at temperatures lower than 35
°C claimed to be the mesophilic optimum (Lin et al.,
1987).
ADM1 contains several biochemical reactions,
physiochemical equilibriums and gas-liquid transfer
processes which all are affected by temperature.
Increased knowledge about AD process parameters
temperature dependency is generally required for
adequate AD modeling using the standard ADM1 at
temperatures other than 35 °C.
Temperature effects on the biogas production is
known (Henze and Harremoës, 1983) but the effects on
individual kinetic coefficients for particle
disintegration, hydrolysis and substrate uptake reactions
might be required to adequately simulate AD with both
particulate and dissolved organics in the substrate where
it is not known which is the rate-limiting reaction.
Temperature dependency for the kinetics in AD is
often described using one Arrhenius equation (Banik et
al., 1998; Grant and Lin 1995; Kettunen and
Rintala,1997) for exponential growth in the mesophilic
area up to 30 °C, but in order to include both the
opposing synthesis and degradation processes the
temperature dependency for the biochemical reactions
in the whole mesophilic range should be modelled using
double Arrhenius equations (1) (Hinshelwood, 1946;
Pavlostathis and Giraldo-Gomez, 1991).
−�� −��
� = �1 x ��
− �2 x (1)
��
ρ = microbial activity, A is the frequency factor, Ea is
the apparent activation energy, R is the gas constant and
T is the absolute temperature.
This temperature effect on each individual
degradation reaction are studied to some degree
(Donoso-Bravo et al., 2009; Rebac et al., 1995, Lin et
al., 1987) but by using varying models for the
temperature effect. Intermediate products are measured
here to evaluate the model’s ability to simulate
individual degradation reaction.
261
 Temperature Effects in Anaerobic Digestion Modeling
It is probably sufficient that the rate limiting
degradation step is modeled utilizing the proper
temperature effect(s). It is however not always obvious
what the rate-limiting step is. Hydrolysis and
disintegration are often assumed rate limiting for
particle rich substrates, such as manure, but this may be
altered by pre-treatment. Particle rich manure filtrate,
for which the rate-limiting step is unknown, is used as
feed in this study.
Modelling AD at thermophilic (45 - 70 °C) and
psychrophilic (4 - 15 °C) temperatures with possible
other degradation paths (Vavilin et al., 1997) is not
attempted here. Temperature effects on diffusion are
assumed insignificant in the range investigated.
An intention is that the model can be used to reveal
optimal sludge bed AD temperatures, such as to
investigate the potential for reducing heating
requirements by reactor operation at temperatures below
35°C.
The purpose of this study is to quantify temperature
effects on AD modelling by establishing and testing an
extended ADM1 (“ADM1-T”) to account for
temperature effects on model parameters.
The model is tested on data from a 220 liter AD
sludge bed reactor treating diary manure filtrate for 4
months of varying loads and with step temperature
changes between 25, 30 and 35 °C.
Additional aims are:
(1) Distinguish between physical and biological
effects of temperature;
(2) Evaluate temperature effects in sludge bed AD
and
(3) Look for limiting reaction steps for process
capacity by testing and modeling AD at different loads
and temperatures.
This should improve our general understanding of
how, where and when temperature influences AD
processes.
2 Materials and Methods
Temperature dependent kinetic parameters for both
biochemical and physico-chemical processes are
retrieved from literature survey. Other relevant model
parameters are retrieved from batch tests and compared
against continuous AD using diary manure filtrate.
2.1 Model Parameters
2.1.1 Biochemical kinetic parameters
Kinetic temperature dependent parameters for
biochemical processes in ADM1-T are Kdis, Khyd, km,
Kd, Y and KS, which are recommended in ADM1 for 35
°C and used as a reference (Table 1). Kdis and Khyd are
for 1.st order extracellular reactions disintegration and
hydrolysis (2).
� = ��� ∙ ��� (2)
,ℎ�� ,ℎ��
262 Proceedings of the 56th SIMS
October 07-09, 2015, Linköping, Sweden
ρ = disintegration rate or hydrolysis rate of solid
substrate (kg COD solid substrate m-3 d-1 where COD =
chemical oxygen demand), Xdis,hyd = solid substrate
concentration (kg COD solid substrate m-3), Kdis,hyd =
temperature dependent kinetic parameter for
disintegration or hydrolysis (d-1). Disintegration is
typically considered the rate-limiting step for substrates
containing mainly particles, while hydrolysis of
proteins, lipids and carbohydrates is the rate limiting in
high rate digesters and then only disintegration of
decaying microorganisms is accounted for (Batstone et
al., 2002).
Figure 1. COD flow diagram of the Anaerobic Digestion
Model No.1 (Adapted from Batstone et al., 2002)
showing the biochemical reactions for which temperature
effects have been included as arrows.
Each intracellular enzyme mediated biochemical
reaction (acidogenesis, acetogenesis, methanogenesis)
(Fig. 1) is generally approximated by a Monod type
saturation function where the reaction rate of substrate
uptake by organism, ρ (kg COD substrate m-3 d-1), can
be described as:
ρ=� ∙X∙ +�
∙I (3)
Equation (3) contains the maximum substrate uptake
rate constant km (kg COD substrate kg COD biomass-1
d-1), X = biomass concentration (kg COD biomass m-3),
S = substrate concentration (kg COD substrate m-3), KS
= half saturation constant (kg COD substrate m-3) and I
= inhibition factor. The growth of biomass, X, is
expressed through the yield, Y (kg COD biomass X kg-
1
COD substrate) of uptake of substrate, while biomass
death is described by Kd (d-1) (Table 1).
Temperature effects on Khyd and km are retrieved
from literature sources. The hydrolysis kinetic
parameter is also tested by a set of laboratory
experiments to find both the khyd for this substrate and
the effect of low temperature. A Hinshelwood double
DOI
10.3384/ecp15119261
 Session 9A: Session A

Arrhenius function is recommended (Batstone et al., degraders at 10 – 40 °C. Calculated relative factors from
2002) as the temperature dependency for the these temperature effects are also in Table 3. Reactions
biochemical reactions expressed through km. The that were not found were estimated by using factors
affinity for the substrate parameter KS is therefore not from “nearby” reactions in Fig. 1: The same temperature
altered. Monod (1949) assumes a saturation relationship effect as for sugar uptake is used on amino acids and
between substrate concentration and growth rate while fatty acids uptake. The same temperature effect for
growth yield remained constant over a wide range of hydrogen uptake as for acetate uptake is also used.
substrate concentrations that requires a constant KS. We
extend this assumption to temperature and evaluate if
temperature effects on yield can be ignored. The death Table 3. Relative change of the kinetic parameters Kdis,
Khyd and km with temperature. Calculated from reported
of biomass Kd has a constant small value in ADM1. The
data in (A) Donoso-Bravo et al. (2009), and (B) Rebac et
temperature effect, if any, is assumed small and not
al. (1995).
altered.
Process Temperature (°C) Ref.
25 30 35
Table 1 Biochemical processes temperature dependent Disintegration, Same as for hydrolysis
kinetic parameters. Kdis of carbohydrates
Parameter Biochemical Expre Denomination Hydrolysis of 0.48 0.74 1.00 A
ssion carbohydrates,
Kdis disintegration 2 d-1 Khyd, su
Khyd hydrolysis 2 d-1 Hydrolysis of Same as for hydrolysis
km max substrate 3 kg COD protein, Khyd, pr of carbohydrates
uptake rate substrate Hydrolysis of lipids, Same as for hydrolysis
constant m-3 d-1 Khyd, li of carbohydrates
KS half saturation 3 kg COD Sugar 0.21 0.22 1.00
constant substrate m-3 A
Uptake, km
Y biomass kg COD Amino acid uptake, Same as for sugar
growth yield biomass km uptake
kg-1 COD Same as for sugar
substrate Fatty acid uptake, km
uptake
Kd death rate of d-1 Butyrate uptake, km 0.67 0.86 1.00 B
biomass Propionate uptake 0.70 0.90 1.00 B
Aceticlastic 0.69 0.93 1.00
A
Relative factors of the relevant temperatures, 35, 30 methanogens, km
and 25 °C are calculated from reported temperature Aceticlastic 0.48 0.70 1.00
B
effect (Henze and Harremoës, 1983) on the overall methanogens, km
biogas production (Table 2). These factors are applied Hydrogenotroph Same as for aceticlastic
for all the kinetic parameters Kdis, Khyd and km for all the methanogens, km methanogens
degradation steps since the limiting reaction is unknown
as a first approach. Three different simulations were done;
H: using the same temperature effect (Table 2) for all
Table 2. Relative temperature factors for kinetic the degradation steps.
parameters for the overall biogas production. Calculated D: using all the temperature effects from Table 3 but
from Henze and Harremoës (1983). with methanogenesis from Donoso-Bravo et al. (2009)
Temperature (°C) 25 30 35
R: using all the temperature effects from Table 3 but
Temperature factor 0.42 0.87 1 with methanogenesis from Rebac et al. (1995)
2.1.2 Physio-chemical parameters
Temperature effects on the kinetic parameters that are
different for each degradation equation are also The temperature dependencies of the temperature
implemented. Relative temperature effects for dependent kinetic parameters ka and kH (Table 4) are in
hydrolysis, acidogenesis and acidoclastic the standard ADM1 while the temperature dependency
methanogenesis of respectively starch, glucose and of the mass transfer coefficient kLa is implemented here.
acetic acid through batch tests at 15-45 °C (Donoso- Physico-chemical equilibrium is modeled based on
Bravo et al., 2009) is used to calculate temperature the law of mass action for aqueous substances and on
factors for each reaction (Table 3). Relative temperature Henry’s law to model the solubility of a gas in water.
effects on individual reactions are also reported by Both are temperature dependent; which are given by
Rebac et al. (1995) for butyrate, propionate and acetate equations (4) and (5) respectively.

DOI Proceedings of the 56th SIMS 263

10.3384/ecp15119261 October 07-09, 2015, Linköping, Sweden
 Temperature Effects in Anaerobic Digestion Modeling

parameter for this substrate. 30 ml of manure filtrate was

�− ∙ �+ added to the syringe and the gas production was
�� = −
(4)
� ��� � monitored regularly. The soluble and total chemical
oxygen demand, CODS and CODT, (representing the
������ = �� ∙ � (5) energy content of the waste) was measured before and
after the test. The not degradable fraction of the
Where both Ka and kH are temperature dependent substrate was calculated based on remaining COD. The
according to van Hoff’s equation (Eq. 6). CODT removed is equivalent to the biogas potential. To
determine how much of the COD came from either the
∆� ° 1 1
�� = ∙ −� (6) soluble (B0soluble) or the solid (B0particle) fractions, a factor
� �
f = B0soluble/B0particle was estimated by COD particle and
liquid contents at start and end of the experiment.
The equilibriums reactions (7-9) have temperature
The hydrolysis kinetic factor Kh was calculated
correction implemented while carboxylic acids (10)
using:
temperature dependency is assumed small and not taken
B = B0 (1-e-kt) (12)
into account (Batstone et al., 2002).
at temperatures 23 and 35 °C. B0 is the total biogas
CO2 + H2O ↔ HCO3- + H+ (7)
production and B is the biogas production at the given
NH4+ + H2O ↔ NH3 + H+ (8)
time t.
H2O ↔ H+ + OH- (9)
R-COOH ↔ R-COO- + H+ (10) 2.3 AD sludge bed experiment
Dynamic gas transfer is also dependent on 2.3.1 AD reactor design operation
temperature, accounted for by the temperature The diary manure feed was from the organic milk
dependency of the overall mass transfer coefficient, kLa producer Foss Farm in Skien, Norway. The cattle were
(Eq.11), according to Saravanan et al. (2000). fed grass and clover ensilage and 25 % “dairy
concentrate” (19 % protein). The manure was flushed
kLa = 0.56 ∙ T + 27.9 (11) into an indoor temporarily storage tank, diluting the
manure with flush water by 14 % on average (Bergland
Table 4. Temperature dependent physico-chemical et al., 2014). The manure was stored for 1-6 weeks in
processes parameters. the storage tank before treated in a rotating vacuum
Parameter Physico- Expression drum filter (mesh light opening of 1.4 mm) to remove
chemical in Equations the coarse solids. The filtrate was used as AD feed
ka Equilibrium (4) (substrate) in this study.
constant The diary manure AD was performed in a sludge bed
kH Henry’s constant (5) reactor as an integral part of a process to generate
kLa Mass transfer (11) fertilizers and biogas, as described by Haugen et al.
coefficient (2013) (Fig. 2).

2.1.3 Non temperature dependent parameters

The conditions with longer sludge retention time (SRT)
than hydraulic retention time (HRT) is implemented in
ADM1 as SRT = tres_x + HRT, recommended with
tres_x = 40 days for high rate, sludge bed reactors. The
validation experiment was carried out in a sludge bed
reactor, but not at high rate, so lower tres_x is also
evaluated.
ADM1 uses the same fixed stoichiometry to model
product distribution for 35 and 55 °C. These
stoichiometric values where assumed appropriate for the
Figure 2. The sludge bed AD process of this pilot plant
range from 25 to 35 °C also.
treating dairy cow manure was the subject of this study.
2.2 Biodegradability and hydrolysis
Long-term batch tests using 100 ml medical syringes as The AD reactor had a liquid volume of 254 liter with
batch reactors were performed on diary manure filtrate 400 mm Ø and 2000 mm height consisting mainly of a
at room temperature (~23 °C) and at 35 °C to determine lower liquid volume for suspended biomass. A plastic
the biodegradable amount and the kinetic hydrolysis support medium for biofilm growth was integrated with
solids, gas and liquid separation at the top. The process

264 Proceedings of the 56th SIMS DOI

October 07-09, 2015, Linköping, Sweden 10.3384/ecp15119261
 Session 9A: Session A

had been operated for 2 year at 35°C following 70 days

at 25 °C before the 130 days operation reported here
allowing the culture to adapt to cow manure filtrate as
substrate.
The reactor was operated at the mesophilic
temperatures 25, 30 and 35 °C during this test. The load
was from zero up to a load corresponding to a hydraulic
retention time (HRT) of 8.5 days. The reactor was semi-
continuously operated by pulse feeding with the feeding
pump controlled as a binary (On/Off) device.
Hydrolysis kinetics were also found from the batch
test that lasted 91 days, excluding data from the first 2-
3 weeks (when hydrolysis is not the rate-limiting step).
Using least square method on Eq.12 (Chap 2.2) gives
estimation of khyd to be 0.13 d-1 at 35 and 0.05 d-1 at 23
°C (Fig. 3). This is in the range of khyd found for straw
and grass tested by Veeken and Hamelers (1999) at
temperatures 20 – 40 °C.
A 250

Accumulated biogas
2.3.2 Monitoring and analysis 200
A comprehensive online- and offline-testing scheme 150
was used to monitor the AD reactor. Biogas production 100

(mL)
(L d-1), gas composition (fractions of CO2 and CH4), 50
liquid flow and reactor temperature were monitored 0
continuously online as described by Haugen et al. 0 50 100 150
(2013). Substrate and effluent samples were collected 1- Time (d)
2 times a week. Total chemical oxygen demand (CODT),
soluble COD (CODS), total solids (TS), volatile solids B
(VS), total suspended solids (TSS), volatile suspended Accumulated biogas 250
solids (VSS), pH, alkalinity, NH4+-N and VFA's 200
(acetate, propionate, butyrate, iso-butyrate, valerate, 150
iso-valerate) were analyzed as described in Bergland et 100
(mL)

al. (2015).
3 Results
3.1 Biodegradability and hydrolysis input to the
model
The batch tests gave a total biodegradable fraction of
0.26 of the CODT. The non-degradable is therefore 74
% for the manure substrate COD used. The fractions of
protein, carbohydrates and lipids in the biodegradable
fraction are set to 0.27 (f_pr), 0.51 (f_ch) and 0.22 (f_li)
respectively (Table 6) as found for slurry cattle manure
(Møller et al., 2003). Tests of the temperature
dependency revealed that 96 % ± 2 % of the biogas
production at 35 °C were achieved at 23 °C in the batch
tests, confirming that yields are quite insensitive to
temperature. The effect is implemented as 96 % at 25
°C and 98 % at 30 °C. The fraction of biodegradable
from particles is below 0.38 and the rest from soluble
organics, determined based on the COD contents before
and after the batch test (Table 5). 0.3 is used as
biodegradable fraction from particles and 0.7 from
dissolved COD in the model.
Table 5. Diary manure filtrate parameters used to
calculate the COD content of the degradable and non-
degradable fractions of particulates and dissolved
organics of the AD reactor substrate.
Property Before After Δ COD
CODT 50 37 13
CODS 14 6 8
CODT - CODS = 36 31
CODparticles
50
0
0 50 100 150
Time (d)
Figure 3. Batch test data and fitted line for 35 °C (A) and
23 °C (B) used to calculate khyd.
The average hydrolysis factor for sludge is khyd =
0.183 at 35 °C (Batstone et al., 2002). The khyd values
for cow manure used here are therefor changed by
multiplying the various coefficients for sludge with the
factor 0.13/0.183 at 35 °C to: khyd_ch = 0.18 from 0.25 d-
1
, khyd_li to 0.07 from 0.1 and khyd_pr to 0.14 from 0.2 d-1
for all the simulations. The khyd values are similarly
adjusted down for the lower temperatures with relative
factors of 1 at 35 °C, 0.74 at 30 °C and 0.47 at 25 °C
which is almost identical to the temperature effect found
by Donoso-Bravo et al. (2009) for carbohydrates (Table
3).
The batch results (Fig. 3) also confirms that
temperature influence AD kinetics much more than
stoichiometry (khyd is reduced by 60 % while yield by 4
% at 23 °C compared to at 35 °C).
3.2 AD sludge bed reactor data input to the
model
The biogas production and effluent concentrations of the
pilot AD are presented with the simulated results in
Figure 5 - 6 during the given load and temperature step
changes (Fig. 4). Some scattering in measured AD
substrate composition may be due to real, uncontrolled,
fluctuations in the influent with seasonal and other

DOI Proceedings of the 56th SIMS 265

10.3384/ecp15119261 October 07-09, 2015, Linköping, Sweden
 Temperature Effects in Anaerobic Digestion Modeling

changes in farm operation. Sampling and measurement
errors are also likely on particle rich manure samples.
Such fluctuations are reduced through the AD reactor,
as expected. The inlet fluctuations are considered noise,
so the inflow concentrations are therefore smoothened
by median values, floating by the amount of samples
indicated after each property, for NH4-N (20), CODT
(6), CODS (4), acetate (20), propionate (20) and butyrate
(20).
Table 6. Substrate inflow content to the AD reactor in the
simulation.
Para Content Formula Denominat
meter ion
and thereby establish model steady state conditions that
correspond to that observed at what is defined as time
zero in this study (Fig. 5 and 6). The CO2 concentration
in the biogas was adjusted by adding a constant inflow
substrate HCO3 level (Table 6). The modeled effluent
pH level was forced to match measured values by
adjusting the addition of a constant concentration of ions
in the inflow substrate. The active reactor biomass
concentrations were also tuned to match steady state
performance, found to add up to a level below 10 g COD
L-1 reactor. This is well below the upper limit of 40 g
COD L-1 reactor for sludge bed AD (Kleerebezem et al.,
2006).
40

Temperature (°C)
35
X_c composite 0 g COD L-1
30
X_pr protein f_pr*0.11* g COD L-1 25
(CODT - CODS) 20
X_li lipid f_li*0.11* g COD L-1 0 50 100 150
(CODT - CODS) Time (d)
X_ch carbo- f_ch*0.11* g COD L-1
hydrates (CODT - CODS) 0,04
Load (m3 d-1)

-1
X_I solid inert 0.86* g COD L 0,03
(CODT - CODS) 0,02
S_I soluble 0.03* g COD L-1 0,01
inert (from (CODT - CODS) 0
solid) 0 50 100 150
-1
CODT total 50.6 ± 2 g COD L Time (d)

CODS soluble 13.9 ± 2 g COD L-1

Figure 4. Temperature (°C) and load (m3 d-1) during the
S_I soluble 0.1 * g COD L-1 experiment.
inert (from (CODs-CODVFA)
liquid)
3.3 Simulation of AD reactor
S_su sugar 0.18*f_ch * g COD L -1 Sludge retention time (SRT), a key factor in sludge bed
(CODs - CODVFA) AD processes is evaluated and quantified first to have a
S_aa amino 0.18*f_pr * g COD L-1 SRT calibrated model to investigate temperature
acids (CODS - CODVFA) effects next.
S_fa long chain 0.18*f_li * g COD L-1
3.3.1 Sludge retention time
fatty acids (CODS - CODVFA)
The process simulation is observed to be highly
S_ac acetic acid 3.3 - 3.6 g COD L-1
dependent on SRT, a parameter that is unknown and
S_pro propionic 1.44 - 1.50 g COD L-1 uncontrolled in most sludge bed reactors such as tested
acid here. The simulated gas production has a good fit to the
S_bu butyric 0.64 - 0.65 g COD L-1 experimental values (Fig. 5 A) independent of SRT
acid modelling approach at steady state. The models ability
to predict process changes, however, was quite different
S_IC HNO3- 0.058 M
depending on SRT. SRT has the largest effect on the
S_IN NH4 + 0.07 ± 0.001 M effluent acetate concentration, with SRT calculated
NH3 from tres_x = 20 being closest to the experimental
values. The recommended way of simulating SRT using
tres_x = 40 (Batstone et al., 2002) has the lowest fit to
The reactor was started more than two years before the experimental values for biogas production but good
the test period used here. Simulations of a preliminary fit to the acetate concentration. The observation that
period (340 days before the results presented here) were SRT influence acetate, the reactant for most of the
used to adjust pH and CO2 concentration of the biogas methane production, especially, suggests that

266 Proceedings of the 56th SIMS DOI

October 07-09, 2015, Linköping, Sweden 10.3384/ecp15119261
 Session 9A: Session A

methanogenesis can be especially sensitive to load
transitions and SRT.
SRT is calculated using tres_x = 15, since it gave the
best correlations, for the remaining simulations to
evaluate the impact of the various temperature corrected
kinetic constants. A lower tres_x, implying lower SRT
than that proposed by Batstone et al. (2002) seem
reasonable for the present case since HRT was higher
than typical for sludge bed AD.
A tres_x 6
tres_x 15
day gives reasonable fit (Fig. 6). Larger model
numerical step size overestimates this initial peak. The
experimental temperature steps lasted for 0.4 and 0.5
days from 25 to 30 °C and 30 to 35 °C, respectively.
The three simulations “H”, “D” and “R” gave the
same biogas production except for “H” at 30 °C which
coincidence with the original ADM1 model. The
original model gave higher biogas production at both 25
°C and 30 °C but similar to the others at 35 °C as
expected (Fig. 6).
tres_x 10
A
tres_x 20
0,3

Biogas production (m3 d-1)

tres_x 40 experiment
0,25
0,3
Biogas production (m3 d-1)

0,2
0,25
0,2 0,15

0,15 0,1
0,1 0,05
0,05 0
0 0 50 100 150
0 50 100 150 Time (d)
Time (d)
B
2
B
Acetate concentration

1,5
3,5
(g COD L-1)
Acetate concentration

3
2,5 1
2
(g COD L-1)

1,5 0,5
1
0,5
0 0
0 50 100 150 0 50 100 150

Time (d) Time (d)

C
Figure 5. Biogas production rate (A) and effluent acetate 0,35
Propionate concentration

concentrations (B) using temperature dependency from 0,3

Henze and Harremoës (1983). Experimental (diamond) 0,25
and simulated (lines).
(g COD L-1)

0,2
0,15
3.3.2 Temperature effects 0,1
0,05
The simulated biogas production and effluent
concentrations (Fig. 6) for the various temperature 0
models shows good fit to the measured values. The 0 50 100 150
various temperature effect parameters (Donoso-Bravo et Time (d)
al., 2009; Rebac et al., 1995; Henze and Harremoës,
1983) differ most in simulating acetate concentrations, Figure 6. Biogas production rate (A) and effluent acetic
but less on the overall biogas production rate. (B) and propionic (C) concentration values temperature
The simulated transient peaks in biogas production dependency using original model, Rebac et al. (1995),
following temperature increases are in the same range as Donoso-Bravo et al. (2009), Henze and Harremoës (1983)
observed (Fig 6). Peak shape depends, however, on with tres_x = 15 d. Experimental (♦) and simulated
numeric also: Step size in the simulation below 0.001 (lines); original , D , H _____, R

DOI Proceedings of the 56th SIMS 267

10.3384/ecp15119261 October 07-09, 2015, Linköping, Sweden
 Temperature Effects in Anaerobic Digestion Modeling
The biogas production, relative to 35 °C, from
simulation and experimentally (Table 7) shows that the
simulated relative temperature change for “H” is less
affected by temperature than the published biogas
production temperature effect used to adjust the kinetic
factors (Table 2). This approach with the same
temperature factor on each biochemical reaction is
however better than the original model, but with less
accuracy than using individually temperature effects for
each reaction as for “R”.
Table 7. Simulation and experimental results on the
temperature effect on biogas production rate.
Model 25 °C 30 °C 35 °C
R 0.79 0.85 1 main deviations between measured and modeled values,
D 0.81 0.86 1 observed in transient acetate accumulations, were
H 0.80 0.95 1
Original model 0.95 0.97 1 microbial substrate uptake rate kinetics accounts for the
AD experiment 0.77 0.92 1 observed steady state gas generation levels.
The tested temperature model parameters using the
simulation “R” is predicting the key intermediate,
acetate, best. Significant deviations are only found in the
simulations of the intermediate product propionate (Fig.
6) with elevated propionate following the load increase.
The predictions are less than half of that observed. Note,
however, that these levels are low and close to detection
limit for propionate. Simulated overall soluble and
particulate organic carbon removal and pH are close to
measured values.
The original ADM1 model with only temperature
correction for physio-chemical reactions did not
simulate the observed temperature effects as well as the
new version made here to also include temperature
effects on the bio-reactions. It did however predict the
transient gas production peak as well as the extended
model, implying that this peak has a physical
explanation. It is explained by reduced solubility of
gasses with increased temperature causing gas release.
The elevated transient acetate suggests that
methanogenesis is the rate-limiting step of the process
investigated here, but the results are not conclusive
except during the transient response to the step load
increase (Fig. 5 - 6). It is normally either the
disintegration, hydrolysis, degradation of propionic acid
or the methanogenesis that is the limiting AD reaction
(Batstone et al., 2002). The temperature effects on
biogas production rate reported by Henze and
Harremoës, (1983) may indicate hydrolysis as the rate-
limiting step and not the methanogenesis for substrates
with high particulates content (such as is the case in
manure). The relatively small fraction of particles
degraded in our experiment may however change this,
possibly making methanogenesis the rate limiting
268 Proceedings of the 56th SIMS
October 07-09, 2015, Linköping, Sweden
process step. This analysis does, therefore, not revealed
which biochemical step is rate-limiting during the
various phases of the experiment. It can, however, be
argued that it is not important to know which step is rate
limiting as long as appropriate temperature corrections
are applied on the kinetic parameters of all the
potentially rate–limiting steps. The reasonably good fit
with experimental observations supports this claim.
4 Conclusion
The calibrated model, using recommended ADM1
parameters with standard temperature effect models,
yields good fit of simulated and measured biogas
production. Simulated overall soluble and particulate
organic carbon removal, pH and intermediate products
accumulation are also close to measured values. The
sensitive both to sludge retention time (SRT) and
temperature effect modeling. Temperature effects on
Implementing individual temperature effects for each
biochemical reaction gave the best fit for both biogas
production and effluent acetate concentration. The
original ADM1 model did not simulate the observed
temperature effect as well as the modified models. All
the modified models confirmed the experimental
observation that temperature has a moderate influence
on steady state biogas production in sludge bed AD (1.6
% reduced production per degree at 35 – 30 °C and 3.4
% per degree at 30 – 25°C). It is not revealed which
biochemical step is rate-limiting but that seems
irrelevant in ADM1 simulations.
The influence of temperature on steady state biogas
production in the sludge bed AD was predictable using
standard modeling parameters and moderate (1.6 % per
degree at 30 – 35 °C and 3.4 % per degree at 25 – 30
°C). This imply that the net energy gain can peak at T <
35 °C in some cases and that such maxima can be found
by the temperature extended ADM1.
Acknowledgements
The project was supported by the Norwegian
Agricultural Agency, Innovation Norway, The Research
Council of Norway, Ministry of Education and Research
and Telemark University College. The authors wish to
thank farmer Knut Vasdal for the good cooperation in
carrying out the experiment and Associate Professor
Finn Haugen for the automatic process monitoring and
control.
References
Banik, G., Viraraghavan, T., Dague, R. (1998). Low
temperature effects on anaerobic microbial kinetic
DOI
10.3384/ecp15119261
 Session 9A: Session A

parameters. Environmental Technology, 19(5), 503- Pavlostathis, S., Giraldo-Gomez, E. (1991). Kinetics of
512. anaerobic treatment: A critical review. Critical
Batstone, D., Keller, J., Angelidaki, I., Kalyuzhnyi, S., Reviews in Environmental Control, 21(5-6), 411-
Pavlostathis, S., Rozzi, A., . . . Vavilin, V. (2002). 490.
Anaerobic Digestion Model No. 1 (ADM1). Tech. Rebac, S., Ruskova, J., Gerbens, S., Vanlier, J., Stams,
rep., IWA Publishing. A., Lettinga, G. (1995). High-rate anaerobic
Bergland, W., Dinamarca, C., Bakke, R. (2014). treatment of waste-water under psychrophilic
Efficient biogas production from the liquid fraction conditions. Journal of Fermentation and
of diary manure. Renewable Energy & Power Bioengineering, 80(5), 499-506.
Quality Journal (RE&PQJ), 12. Saravanan, V., Hemachandran, B., Raj, A., Sundaram,
Bergland, W., Dinamarca, C., Toradzadegan, M., S. (2000). Liquid phase volumetric mass transfer
Nordgård, A., Bakke, I., Bakke, R. (2015). High rate coefficient in dairy effluent stream. Bioprocess
manure supernatant digestion. Water Research, 76, Engineering, 23(2), 175-176.
1-9. Vavilin, V., Lokshina, L., Rytov, S., Kotsyurbenko, O.,
Donoso-Bravo, A., Retamal, C., Carballa, M., Ruiz- Nozhevnikova, A., Parshina, S. (1997). Modelling
Filippi, G., Chamy, R. (2009). Influence of methanogenesis during anaerobic conversion of
temperature on the hydrolysis, acidogenesis and complex organic matter at low temperatures. Water
methanogenesis in mesophilic anaerobic digestion: Science and Technology, 36(6-7), 531-538.
parameter identification and modeling application. Veeken, A., Hamelers, B. (1999). Effect of temperature
Water Science and Technology, 60(1), 9-17. on hydrolysis rates of selected biowaste components.
Grant, S., Lin, K. (1995). Effects of temperature and Bioresource Technology, 69(3), 249-254.
organic loading on the performance of upflow
anaerobic sludge blanket reactors. Canadian Journal
of Civil Engineering, 22(1), 143-149.
Hafner, S., Bisogni, J. (2009). Modeling of ammonia
speciation in anaerobic digesters. Water Research,
43(17), 4105-4114.
Haugen, F., Bakke, R., Lie, B. (2013). Adapting
Dynamic Mathematical Models to a Pilot Anaerobic
Digestion Reactor. Modeling, Identification and
Control, 34(2), 35-54.
Henze, M., Harremoes, P. (1983). Anaerobic treatment
of wastewater in fixed film reactors - A litterature
review. Water Science and Technology, 15(8-9), 1-
101.
Hinshelwood, C. (1947). The Chemical Kinetics of the
Bacterial Cell. London: Oxford University Press.
Kettunen, R., Rintala, J. (1997). The effect of low
temperature (5-29 degrees C) and adaptation on the
methanogenic activity of biomass. Applied
Microbiology and Biotechnology, 48(4), 570-576.
Kleerebezem, R., van Loosdrecht, M. (2006). Critical
analysis of some concepts proposed in ADM1. Water
Science and Technology, 54(4), 51-57.
Lin, C., Noike, T., Sato, K., Matsumoto, J. (1987).
Temperature characteristics of the methanogenesis
process in anaerobic-digestion. Water Science and
Technology, 19(1-2), 299-310.
Monod, J. (1949). The Growth of Bacterial Cultures.
Annual Review of Microbiology, 3, 371-394.
Møller, H., Sommer, S., Ahring, B. (2004). Methane
productivity of manure, straw and solid fractions of
manure. Biomass & Bioenergy, 26(5), 485-495.

DOI Proceedings of the 56th SIMS 269

10.3384/ecp15119261 October 07-09, 2015, Linköping, Sweden